Improvised Optical Flow Techniques to Track Vehicle Movements in a Drone Video—Hybrid Approach Combining Lucas–Kanade and Horn–Schunck Methods

This paper focuses on the crucial task of accurately tracking vehicle movements in drone videos for applications such as traffic monitoring and surveillance. Optical flow techniques have shown promise in estimating motion between consecutive frames. In this study, a hybrid approach is adapted that combines the strengths of the Lucas–Kanade method and the Horn–Schunck method to achieve precise tracking of short vehicle movements in drone video frames. Proposed approach aims to improve the precision and efficiency of vehicle motion estimation, thereby contributing to the advancement of computer vision-based vehicle speed estimation in aerial videos. To achieve this, Lucas–Kanade algorithm is being used in two iterations for sparse key points. In between these iterations, Horn–Schunck technique is being employed for intense optical flow estimation. By leveraging modified optical flow technique in proposed novel sandwiched arrangement, the challenges in accurately capturing small vehicle movements are being addressed which are vital for robust vehicle speed estimation, providing improvised results. A detailed analysis showcases the effectiveness of this approach in achieving more reliable motion estimation and precise vehicle speed calculations. This research offers valuable insights into the integration of optical flow techniques for optimizing vehicle motion estimation in drone videos, opening new possibilities for applications in traffic monitoring and aerial surveillance.